Normalization of interval symbolic data

Abstract

As a new tool in data mining, symbolic data analysis (SDA) can not only decrease the computational complexity of huge data sets, but also master the property of the sample integrally. In many statistical analysis methods the sample data need to be normalized in advance. This paper focuses on the normalization of interval symbolic data. Firstly, on the assumption that all the individuals are uniformly distributed within the interval, the method of normalization of interval symbolic data is presented. Then, the method of normalization of interval symbolic data of arbitrary distribution is studied. For this purpose, the mean and variance of interval symbolic data is brought forward in advance. Finally, an example of evaluation on different financial stocks in Chinese stock market is given.